Instance Segmentation by Semi-Supervised Learning and Image Synthesis

Takeru OBA; Norimichi UKITA

doi:10.1587/transinf.2019MVP0016

Instance Segmentation by Semi-Supervised Learning and Image Synthesis

Takeru OBA, Norimichi UKITA

Full Text Views

0

Cite this

Summary :

This paper proposes a method to create various training images for instance segmentation in a semi-supervised manner. In our proposed learning scheme, a few 3D CG models of target objects and a large number of images retrieved by keywords from the Internet are employed for initial model training and model update, respectively. Instance segmentation requires pixel-level annotations as well as object class labels in all training images. A possible solution to reduce a huge annotation cost is to use synthesized images as training images. While image synthesis using a 3D CG simulator can generate the annotations automatically, it is difficult to prepare a variety of 3D object models for the simulator. One more possible solution is semi-supervised learning. Semi-supervised learning such as self-training uses a small set of supervised data and a huge number of unsupervised data. The supervised images are given by the 3D CG simulator in our method. From the unsupervised images, we have to select only correctly-detected annotations. For selecting the correctly-detected annotations, we propose to quantify the reliability of each detected annotation based on its silhouette as well as its textures. Experimental results demonstrate that the proposed method can generate more various images for improving instance segmentation.

Publication: IEICE TRANSACTIONS on Information Vol.E103-D No.6 pp.1247-1256

Publication Date: 2020/06/01

Publicized: 2020/03/18

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2019MVP0016

Type of Manuscript: Special Section PAPER (Special Section on Machine Vision and its Applications)

Category

Authors

Takeru OBA
Toyota Technological Institute
Norimichi UKITA
Toyota Technological Institute

Keyword

instance segmentation, semi-supervised learning, 3D CG models, image synthesis

Cite this

Copy

Takeru OBA, Norimichi UKITA, "Instance Segmentation by Semi-Supervised Learning and Image Synthesis" in IEICE TRANSACTIONS on Information, vol. E103-D, no. 6, pp. 1247-1256, June 2020, doi: 10.1587/transinf.2019MVP0016.
Abstract: This paper proposes a method to create various training images for instance segmentation in a semi-supervised manner. In our proposed learning scheme, a few 3D CG models of target objects and a large number of images retrieved by keywords from the Internet are employed for initial model training and model update, respectively. Instance segmentation requires pixel-level annotations as well as object class labels in all training images. A possible solution to reduce a huge annotation cost is to use synthesized images as training images. While image synthesis using a 3D CG simulator can generate the annotations automatically, it is difficult to prepare a variety of 3D object models for the simulator. One more possible solution is semi-supervised learning. Semi-supervised learning such as self-training uses a small set of supervised data and a huge number of unsupervised data. The supervised images are given by the 3D CG simulator in our method. From the unsupervised images, we have to select only correctly-detected annotations. For selecting the correctly-detected annotations, we propose to quantify the reliability of each detected annotation based on its silhouette as well as its textures. Experimental results demonstrate that the proposed method can generate more various images for improving instance segmentation.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019MVP0016/_p

Copy

@ARTICLE{e103-d_6_1247,
author={Takeru OBA, Norimichi UKITA, },
journal={IEICE TRANSACTIONS on Information},
title={Instance Segmentation by Semi-Supervised Learning and Image Synthesis},
year={2020},
volume={E103-D},
number={6},
pages={1247-1256},
abstract={This paper proposes a method to create various training images for instance segmentation in a semi-supervised manner. In our proposed learning scheme, a few 3D CG models of target objects and a large number of images retrieved by keywords from the Internet are employed for initial model training and model update, respectively. Instance segmentation requires pixel-level annotations as well as object class labels in all training images. A possible solution to reduce a huge annotation cost is to use synthesized images as training images. While image synthesis using a 3D CG simulator can generate the annotations automatically, it is difficult to prepare a variety of 3D object models for the simulator. One more possible solution is semi-supervised learning. Semi-supervised learning such as self-training uses a small set of supervised data and a huge number of unsupervised data. The supervised images are given by the 3D CG simulator in our method. From the unsupervised images, we have to select only correctly-detected annotations. For selecting the correctly-detected annotations, we propose to quantify the reliability of each detected annotation based on its silhouette as well as its textures. Experimental results demonstrate that the proposed method can generate more various images for improving instance segmentation.},
keywords={},
doi={10.1587/transinf.2019MVP0016},
ISSN={1745-1361},
month={June},}

Copy

TY - JOUR
TI - Instance Segmentation by Semi-Supervised Learning and Image Synthesis
T2 - IEICE TRANSACTIONS on Information
SP - 1247
EP - 1256
AU - Takeru OBA
AU - Norimichi UKITA
PY - 2020
DO - 10.1587/transinf.2019MVP0016
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 6
JA - IEICE TRANSACTIONS on Information
Y1 - June 2020
AB - This paper proposes a method to create various training images for instance segmentation in a semi-supervised manner. In our proposed learning scheme, a few 3D CG models of target objects and a large number of images retrieved by keywords from the Internet are employed for initial model training and model update, respectively. Instance segmentation requires pixel-level annotations as well as object class labels in all training images. A possible solution to reduce a huge annotation cost is to use synthesized images as training images. While image synthesis using a 3D CG simulator can generate the annotations automatically, it is difficult to prepare a variety of 3D object models for the simulator. One more possible solution is semi-supervised learning. Semi-supervised learning such as self-training uses a small set of supervised data and a huge number of unsupervised data. The supervised images are given by the 3D CG simulator in our method. From the unsupervised images, we have to select only correctly-detected annotations. For selecting the correctly-detected annotations, we propose to quantify the reliability of each detected annotation based on its silhouette as well as its textures. Experimental results demonstrate that the proposed method can generate more various images for improving instance segmentation.
ER -